Skyread outlook: how it is made and how it has done
Skyread forecasts a few dozen prices, pay rates and sales that grocery, restaurant and freight businesses plan around, 3 to 12 months ahead. Every forecast is tested on years of past data first, and an outlook is shown only where its test errors were smaller than simple rules' and its ranges held about as often as they claim. Everything below can be checked.
See also the public accuracy test of Skyread's published data.
Updated 2026-10-31 03:05 UTC, after the latest data update.
Raw results: JSON. Code: forecast.py, forecast_engine.py, evals/forecast_backtest.py.
Reading an outlook
Each outlook gives a central estimate first, then a likely range. The likely range is Skyread's 80% range: in testing, 81% of the published series' 6-month outcomes (1160 test forecasts) fell inside such ranges, about 8 in 10. So about one time in five the outcome lands outside it, more often when the economy changes abruptly.
Prices in dollars, such as diesel and hourly pay, are given in dollars with the change from a year earlier; price indexes (CPI, PPI) as the change from a year earlier only. A series whose current 6-month likely range spans more than 10 points of yearly change gets no outlook, even when it passed every test: a range that wide is too volatile to be useful for a decision. That limit is Skyread's product rule, not a statistical test.
How the outlook is made
Equal-weight average of three standard models (exponential smoothing with a damped trend, ARIMA chosen by AICc from a small fixed set, and Theta), fitted to the log of the monthly level. Ranges come from the average's own past out-of-sample errors at each horizon (split-conformal), not from model formulas. Tested by rolling origin: a forecast from every month since 2008 using only data available then; accuracy and range coverage are measured from 2014 on, including 2020-2023. Published only for series whose test errors were smaller than 'same as a year earlier' at 6 and 12 months and than 'this year's rate of change continues' at 12 months (on average over the test months; at 12 months not a statistically conclusive margin), whose 6-month advantage over both rules is statistically significant (Diebold-Mariano test, one-sided p below 0.10), whose 6-month ranges score better than the year-earlier rule's own ranges, and whose 80% ranges held the outcome 75-85% of the time. A series whose current 6-month range spans more than 10 points of yearly change is withheld as too volatile to be useful. Tests use today's published values; producer prices, pay and sales are revised after release, so their real-time accuracy may be slightly worse. The published series were chosen by these same tests, so their future accuracy is more likely to be a little worse than tested than better.
How it is tested
- Forecasts are of the change from the same month a year earlier, the number Skyread's cards show. The models work on the log of the level, so a range on the level is exactly a range on the yearly change.
- Ranges are set by how wrong the forecasts have actually been: the 80% range is where 80% of past test errors at that horizon fell. Only errors known at the time count, and how often the range held is measured on later months.
- Every test forecast from 2014-01 on is scored, including 2020-03 to 2023-12 (the pandemic and the price surge), which is also shown on its own.
- Tests use the values as published today. Consumer prices are not revised after release, but producer prices (after four months), pay and sales are, so their real-time accuracy may be slightly worse than shown.
- The series shown with an outlook are the ones that passed these same tests, so their future accuracy is more likely to be a little worse than their test results than better.
- How often the 12-month ranges held differed a lot between 2020–23 and other years (see the last two columns): the overall share is an average over both, so in calm years they are often wider than needed, and in an abrupt change they miss more often.
- Weekly fuel prices are forecast as monthly averages. Where USDA's Food Price Outlook covers the same price index, both are compared on the same yearly-average measure.
When an outlook is published
All of these, over every test forecast, checked again each time new data arrives. Each rule was written down before it was scored (the significance, range-score and usefulness rules in evals/results/forecast_v2_registration.md).
- More accurate than 'same as a year earlier' (the seasonal naive rule) at 6 and at 12 months: relative error below 1 (the forecast's mean absolute error divided by the rule's, on the same test months).
- More accurate than 'this year's rate of change continues' at 12 months: relative error below 1.
- At 6 months, more accurate than both rules by a statistical test: Diebold-Mariano, one-sided, with the Harvey-Leybourne-Newbold correction, p below 0.10 against each rule.
- At 6 months, its likely ranges score better than ranges built the same way around 'same as a year earlier' (the interval score of Gneiting & Raftery, which rewards narrow ranges and penalises misses).
- The 80% ranges held the outcome 75% to 85% of the time in testing, pooled over 3, 6 and 12 months.
- At least 60 test forecasts at 6 and at 12 months.
- Usefulness (Skyread's product rule, not a statistical test): the current 6-month likely range spans at most 10 points of yearly change. A wider one is too volatile to call usefully, so no outlook is given.
Reading the error figures
Relative error is the forecast's average error divided by the simple rule's, on the same months: 0.700 means 30% smaller errors; 1 or more means the simple rule did as well or better. The significance figure is the p-value of a one-sided Diebold-Mariano test with the Harvey-Leybourne-Newbold correction: below 0.10 means an advantage that large would arise by chance less than one time in ten if Skyread were no better than the rule. The range score is the mean interval score of Skyread's 80% ranges divided by that of ranges built the same way around 'same as a year earlier': below 1 means narrower ranges, or fewer and smaller misses. These are averages over the test months: the 12-month figures rest on only about a dozen non-overlapping years since 2014, so a value just under 1 means about as accurate as the simple rule, and even larger 12-month gains are not statistically conclusive. Briefings, the home page and assistants quote the 6-month outlook; the 12-month one is shown here and in the full series answer, with this note: At 12 months, our tests could not show these outlooks are reliably better than simple rules.
Backtest by series
Test forecasts from 2014-01; 12-month figures unless marked. Lower relative error is better.
| Series | Outlook | vs same as a year earlier, 6 / 12 mo | vs rate continues | Significance at 6 mo (p vs year-earlier / vs rate continues) | Range score vs year-earlier rule, 6 mo | Current 6-month likely range (pts) | Error in the yearly change (pts): Skyread / year-earlier rule | 80% range held, 3 / 6 / 12 mo (pooled) | 2020–23: relative error / 80% held | Other years |
|---|---|---|---|---|---|---|---|---|---|---|
| Grocery prices bls.cpi.food_at_home · data through 2026-09 |
Published | 0.420 / 0.760 | 0.764 | 0.004 / 0.060 | 0.519 | 4.6 | 2.2 / 2.8 | 86% / 77% / 74% (79%) | 0.744 / 37% | 0.797 / 91% |
| Grocery meat, poultry, fish and egg prices bls.cpi.meats_poultry_fish_eggs · data through 2026-09 |
Published | 0.524 / 0.893 | 0.692 | 0.002 / 0.002 | 0.653 | 7.0 | 3.6 / 4.0 | 80% / 79% / 78% (79%) | 0.882 / 61% | 0.904 / 86% |
| Beef prices at the grocery store bls.cpi.beef · data through 2026-09 |
Too volatile | 0.558 / 0.886 | 0.755 | 0.002 / 0.011 | 0.715 | 11.6 | 5.9 / 6.6 | 78% / 77% / 71% (75%) | 0.953 / 65% | 0.840 / 74% |
| Pork prices at the grocery store bls.cpi.pork · data through 2026-09 |
Published | 0.614 / 0.964 | 0.675 | 0.008 / 0.008 | 0.653 | 9.0 | 3.7 / 3.9 | 81% / 86% / 84% (84%) | 0.864 / 76% | 1.081 / 87% |
| Chicken and other poultry prices at the grocery store bls.cpi.poultry · data through 2026-09 |
Withheld | 0.627 / 0.966 | 0.800 | 0.031 / 0.099 | 0.695 | 5.9 | 3.2 / 3.3 | 74% / 73% / 71% (73%) | 0.893 / 33% | 1.174 / 89% |
| Egg prices at the grocery store bls.cpi.eggs · data through 2026-09 |
Withheld | 0.722 / 1.067 | 0.660 | 0.014 / <0.001 | 0.717 | 37.8 | 20.1 / 18.8 | 61% / 56% / 52% (57%) | 1.062 / 59% | 1.070 / 49% |
| Dairy prices at the grocery store bls.cpi.dairy · data through 2026-09 |
Withheld | 0.548 / 1.022 | 0.655 | 0.031 / 0.018 | 0.567 | 5.5 | 2.8 / 2.7 | 88% / 91% / 91% (90%) | 0.849 / 72% | 1.391 / 100% |
| Fruit and vegetable prices at the grocery store bls.cpi.fruits_vegetables · data through 2026-09 |
Withheld | 0.678 / 0.848 | 0.713 | 0.034 / 0.031 | 0.722 | 6.0 | 2.0 / 2.4 | 91% / 88% / 80% (87%) | 0.700 / 54% | 1.047 / 93% |
| Bread, cereal and bakery prices at the grocery store bls.cpi.cereals_bakery · data through 2026-09 |
Published | 0.436 / 0.802 | 0.735 | 0.030 / 0.063 | 0.460 | 4.1 | 2.3 / 2.9 | 80% / 81% / 81% (81%) | 0.730 / 48% | 1.039 / 98% |
| Prices of coffee, soft drinks and other nonalcoholic drinks at the grocery store bls.cpi.nonalcoholic_beverages · data through 2026-09 |
Published | 0.476 / 0.780 | 0.843 | 0.005 / 0.050 | 0.567 | 4.7 | 2.6 / 3.3 | 82% / 82% / 76% (80%) | 0.800 / 30% | 0.746 / 99% |
| Coffee prices at the grocery store bls.cpi.coffee · data through 2026-09 |
Published | 0.543 / 0.942 | 0.745 | 0.019 / 0.002 | 0.522 | 9.4 | 4.6 / 4.9 | 86% / 83% / 81% (83%) | 0.898 / 65% | 0.970 / 88% |
| Sugar and sweets prices at the grocery store bls.cpi.sugar_sweets · data through 2026-09 |
Withheld | 0.393 / 0.691 | 0.870 | <0.001 / 0.033 | 0.547 | 5.1 | 2.7 / 3.9 | 86% / 88% / 83% (85%) | 0.635 / 57% | 0.757 / 96% |
| Butter, oil and other fats prices at the grocery store bls.cpi.fats_oils · data through 2026-09 |
Withheld | 0.574 / 0.931 | 0.800 | 0.063 / 0.102 | 0.550 | 7.8 | 3.5 / 3.7 | 85% / 81% / 84% (84%) | 0.789 / 61% | 1.416 / 96% |
| Fish and seafood prices at the grocery store bls.cpi.fish_seafood · data through 2026-09 |
Published | 0.543 / 0.945 | 0.771 | 0.003 / 0.005 | 0.603 | 6.4 | 3.0 / 3.2 | 79% / 77% / 71% (76%) | 0.831 / 54% | 1.073 / 79% |
| Prices of snacks, condiments, baby food and prepared foods at the grocery store bls.cpi.other_food_at_home · data through 2026-09 |
Published | 0.411 / 0.748 | 0.726 | 0.017 / 0.038 | 0.502 | 4.4 | 2.1 / 2.8 | 84% / 81% / 81% (82%) | 0.735 / 48% | 0.795 / 98% |
| Restaurant menu prices bls.cpi.food_away · data through 2026-09 |
Withheld | 0.144 / 0.289 | 1.152 | <0.001 / 0.517 | 0.364 | 1.7 | 1.2 / 4.0 | 78% / 79% / 69% (76%) | 0.399 / 22% | 0.191 / 93% |
| Fast-food and counter-service menu prices bls.cpi.limited_service_meals · data through 2026-09 |
Withheld | 0.163 / 0.297 | 1.293 | <0.001 / 0.682 | 0.424 | 2.1 | 1.3 / 4.2 | 83% / 81% / 79% (81%) | 0.424 / 37% | 0.169 / 100% |
| Sit-down restaurant menu prices bls.cpi.full_service_meals · data through 2026-09 |
Withheld | 0.154 / 0.302 | 0.968 | <0.001 / 0.202 | 0.372 | 1.5 | 1.2 / 3.9 | 71% / 72% / 73% (72%) | 0.404 / 37% | 0.211 / 90% |
| Food manufacturers' selling prices bls.ppi.food_manufacturing · data through 2026-09 |
Withheld | 0.560 / 0.917 | 0.735 | 0.009 / 0.109 | 0.620 | 7.0 | 3.4 / 3.7 | 87% / 84% / 79% (83%) | 0.915 / 54% | 0.919 / 91% |
| Wholesale prices of processed foods and feeds bls.ppi.processed_foods · data through 2026-09 |
Withheld | 0.567 / 0.926 | 0.748 | 0.014 / 0.126 | 0.601 | 6.5 | 3.1 / 3.4 | 83% / 82% / 77% (81%) | 0.916 / 54% | 0.941 / 87% |
| Farm prices bls.ppi.farm_products · data through 2026-09 |
Withheld | 0.730 / 1.014 | 0.762 | 0.052 / 0.052 | 0.686 | 21.5 | 9.9 / 9.8 | 79% / 79% / 84% (81%) | 0.977 / 50% | 1.062 / 100% |
| Wholesale beef prices bls.ppi.beef · data through 2026-09 |
Too volatile | 0.742 / 0.999 | 0.626 | 0.025 / 0.005 | 0.778 | 24.3 | 12.6 / 12.6 | 81% / 80% / 74% (78%) | 1.024 / 61% | 0.974 / 81% |
| Wholesale chicken prices bls.ppi.chicken · data through 2026-09 |
Withheld | 0.719 / 1.036 | 0.622 | 0.057 / 0.002 | 0.700 | 17.9 | 9.6 / 9.3 | 75% / 70% / 65% (70%) | 0.983 / 41% | 1.092 / 76% |
| Wholesale cheese prices bls.ppi.cheese · data through 2026-09 |
Withheld | 0.706 / 1.044 | 0.571 | 0.003 / <0.001 | 0.834 | 18.4 | 7.4 / 7.1 | 88% / 90% / 97% (92%) | 1.023 / 93% | 1.058 / 99% |
| Wholesale wheat flour prices bls.ppi.wheat_flour · data through 2026-09 |
Withheld | 0.623 / 1.053 | 0.793 | 0.010 / 0.026 | 0.577 | 21.0 | 8.5 / 8.1 | 90% / 90% / 89% (90%) | 0.947 / 65% | 1.174 / 100% |
| Truckload freight rates bls.ppi.truckload · data through 2026-09 |
Withheld | 0.648 / 0.990 | 0.775 | 0.011 / 0.133 | 0.638 | 17.0 | 8.0 / 8.1 | 65% / 60% / 56% (60%) | 0.944 / 26% | 1.050 / 71% |
| Less-than-truckload freight rates bls.ppi.ltl · data through 2026-09 |
Withheld | 0.540 / 0.785 | 0.787 | <0.001 / 0.058 | 0.713 | 10.5 | 5.1 / 6.4 | 75% / 68% / 75% (73%) | 0.923 / 46% | 0.689 / 89% |
| Trucking rates bls.ppi.truck_transportation · data through 2026-09 |
Withheld | 0.585 / 0.968 | 0.754 | 0.009 / 0.090 | 0.579 | 10.3 | 5.6 / 5.8 | 63% / 62% / 56% (60%) | 0.940 / 22% | 1.007 / 73% |
| Warehousing and storage prices bls.ppi.warehousing · data through 2026-09 |
Withheld | 0.646 / 0.827 | 0.850 | 0.031 / 0.469 | 0.677 | 8.3 | 4.0 / 4.8 | 59% / 52% / 58% (56%) | 0.806 / 41% | 0.860 / 66% |
| Diesel at the pump eia.diesel_price · data through 2026-10 |
Too volatile | 0.613 / 0.996 | 0.767 | <0.001 / 0.043 | 0.695 | 48.1 | 21.6 / 21.6 | 81% / 80% / 77% (80%) | 0.990 / 50% | 1.002 / 91% |
| Gasoline at the pump eia.gasoline_price · data through 2026-10 |
Withheld | 0.714 / 1.029 | 0.732 | 0.005 / 0.045 | 0.788 | 45.0 | 17.7 / 17.2 | 81% / 82% / 84% (82%) | 1.008 / 57% | 1.047 / 97% |
| Restaurant hourly pay bls.ces.food_services_ahe · data through 2026-08 |
Withheld | 0.255 / 0.467 | 1.190 | <0.001 / 0.550 | 0.497 | 2.0 | 2.3 / 4.9 | 72% / 69% / 71% (71%) | 0.586 / 48% | 0.365 / 82% |
| Grocery store hourly pay bls.ces.grocery_ahe · data through 2026-08 |
Withheld | 0.453 / 0.691 | 0.886 | <0.001 / 0.015 | 0.643 | 5.3 | 3.1 / 4.5 | 76% / 72% / 66% (71%) | 0.745 / 50% | 0.625 / 74% |
| Trucking hourly pay bls.ces.truck_transport_ahe · data through 2026-08 |
Withheld | 0.338 / 0.582 | 1.141 | <0.001 / 0.460 | 0.556 | 3.2 | 2.1 / 3.5 | 77% / 72% / 79% (76%) | 0.581 / 65% | 0.583 / 87% |
| Warehouse hourly pay bls.ces.warehousing_ahe · data through 2026-08 |
Withheld | 0.467 / 0.693 | 0.789 | <0.001 / 0.035 | 0.768 | 3.7 | 2.3 / 3.3 | 74% / 73% / 65% (71%) | 0.785 / 35% | 0.629 / 81% |
| Restaurant and bar sales census.food_services_sales · data through 2026-09 |
Withheld | 0.510 / 0.768 | 0.647 | <0.001 / 0.047 | 0.697 | 7.0 | 7.9 / 10.2 | 70% / 67% / 66% (68%) | 0.970 / 22% | 0.396 / 87% |
| Grocery store sales census.grocery_store_sales · data through 2026-09 |
Withheld | 0.448 / 0.623 | 0.708 | <0.001 / 0.006 | 0.814 | 4.3 | 2.5 / 4.0 | 71% / 74% / 72% (72%) | 0.745 / 26% | 0.473 / 94% |
The "about 8 in 10" above is pooled over all published series. Where one series' own record differs:
- Grocery prices: In testing, this series' 6-month likely range held 77% of the time, a little below the 80% aim. Outcomes landed outside it a little more often than the range implies.
- Pork prices at the grocery store: In testing, this series' 6-month likely range held 86% of the time, above the 80% aim. The range is wider than it needs to be.
- Prices of coffee, soft drinks and other nonalcoholic drinks at the grocery store: In testing, this series' 6-month likely range held 82.1% of the time, a little above the 80% aim. The range is a little wider than it needs to be.
- Coffee prices at the grocery store: In testing, this series' 6-month likely range held 83% of the time, a little above the 80% aim. The range is a little wider than it needs to be.
- Fish and seafood prices at the grocery store: In testing, this series' 6-month likely range held 77% of the time, a little below the 80% aim. Outcomes landed outside it a little more often than the range implies.
Withheld, and why
No outlook is shown for these series; their published data is unaffected.
- Beef prices at the grocery store: too volatile to call usefully 6 months out: the 80% range would span 11.6 points of yearly change (Skyread publishes a 6-month outlook only when its range spans 10 points or less).
- Chicken and other poultry prices at the grocery store: 80% ranges held the outcome 73% of the time in testing (must be 75-85%).
- Egg prices at the grocery store: not more accurate than 'same as a year earlier' at 12 months (relative error 1.0675); 80% ranges held the outcome 57% of the time in testing (must be 75-85%).
- Dairy prices at the grocery store: not more accurate than 'same as a year earlier' at 12 months (relative error 1.0221); 80% ranges held the outcome 90% of the time in testing (must be 75-85%).
- Fruit and vegetable prices at the grocery store: 80% ranges held the outcome 87% of the time in testing (must be 75-85%).
- Sugar and sweets prices at the grocery store: 80% ranges held the outcome 85% of the time in testing (must be 75-85%).
- Butter, oil and other fats prices at the grocery store: at 6 months, not shown to be more accurate than 'this year's rate of change continues' (Diebold-Mariano one-sided p 0.102; below 0.10 needed).
- Restaurant menu prices: not more accurate than 'this year's rate of change continues' at 12 months (relative error 1.1518); at 6 months, not shown to be more accurate than 'this year's rate of change continues' (Diebold-Mariano one-sided p 0.517; below 0.10 needed).
- Fast-food and counter-service menu prices: not more accurate than 'this year's rate of change continues' at 12 months (relative error 1.2931); at 6 months, not shown to be more accurate than 'this year's rate of change continues' (Diebold-Mariano one-sided p 0.682; below 0.10 needed).
- Sit-down restaurant menu prices: 80% ranges held the outcome 72% of the time in testing (must be 75-85%); at 6 months, not shown to be more accurate than 'this year's rate of change continues' (Diebold-Mariano one-sided p 0.202; below 0.10 needed).
- Food manufacturers' selling prices: at 6 months, not shown to be more accurate than 'this year's rate of change continues' (Diebold-Mariano one-sided p 0.109; below 0.10 needed).
- Wholesale prices of processed foods and feeds: at 6 months, not shown to be more accurate than 'this year's rate of change continues' (Diebold-Mariano one-sided p 0.126; below 0.10 needed).
- Farm prices: not more accurate than 'same as a year earlier' at 12 months (relative error 1.0138).
- Wholesale beef prices: too volatile to call usefully 6 months out: the 80% range would span 24.3 points of yearly change (Skyread publishes a 6-month outlook only when its range spans 10 points or less).
- Wholesale chicken prices: not more accurate than 'same as a year earlier' at 12 months (relative error 1.0359); 80% ranges held the outcome 70% of the time in testing (must be 75-85%).
- Wholesale cheese prices: not more accurate than 'same as a year earlier' at 12 months (relative error 1.0437); 80% ranges held the outcome 92% of the time in testing (must be 75-85%).
- Wholesale wheat flour prices: not more accurate than 'same as a year earlier' at 12 months (relative error 1.0532); 80% ranges held the outcome 90% of the time in testing (must be 75-85%).
- Truckload freight rates: 80% ranges held the outcome 60% of the time in testing (must be 75-85%); at 6 months, not shown to be more accurate than 'this year's rate of change continues' (Diebold-Mariano one-sided p 0.133; below 0.10 needed).
- Less-than-truckload freight rates: 80% ranges held the outcome 73% of the time in testing (must be 75-85%).
- Trucking rates: 80% ranges held the outcome 60% of the time in testing (must be 75-85%).
- Warehousing and storage prices: 80% ranges held the outcome 56% of the time in testing (must be 75-85%); at 6 months, not shown to be more accurate than 'this year's rate of change continues' (Diebold-Mariano one-sided p 0.469; below 0.10 needed).
- Diesel at the pump: too volatile to call usefully 6 months out: the 80% range would span 48.1 points of yearly change (Skyread publishes a 6-month outlook only when its range spans 10 points or less).
- Gasoline at the pump: not more accurate than 'same as a year earlier' at 12 months (relative error 1.0288).
- Restaurant hourly pay: not more accurate than 'this year's rate of change continues' at 12 months (relative error 1.1896); 80% ranges held the outcome 71% of the time in testing (must be 75-85%); at 6 months, not shown to be more accurate than 'this year's rate of change continues' (Diebold-Mariano one-sided p 0.550; below 0.10 needed).
- Grocery store hourly pay: 80% ranges held the outcome 71% of the time in testing (must be 75-85%).
- Trucking hourly pay: not more accurate than 'this year's rate of change continues' at 12 months (relative error 1.1413); at 6 months, not shown to be more accurate than 'this year's rate of change continues' (Diebold-Mariano one-sided p 0.460; below 0.10 needed).
- Warehouse hourly pay: 80% ranges held the outcome 71% of the time in testing (must be 75-85%).
- Restaurant and bar sales: 80% ranges held the outcome 68% of the time in testing (must be 75-85%).
- Grocery store sales: 80% ranges held the outcome 72% of the time in testing (must be 75-85%).
Against USDA's Food Price Outlook
In our backtest of USDA ERS's 2023–2025 Food Price Outlook editions, using only the CPI data ERS had at each edition, Skyread's errors were smaller in 12 of 14 categories. That covers three years of outcomes, and our method was chosen in 2026 while ERS forecast in real time. ERS's errors were smaller for egg prices at the grocery store, fish and seafood prices at the grocery store.
USDA ERS forecasts the change in each food category's average price over the calendar year. For each ERS edition from 2023 to 2025 that forecast its own year, Skyread's backtest forecast made with the same CPI month ERS had (the month before the edition) is turned into the same yearly-average change, and both are compared with what BLS later published. Error in percentage points; lower is better.
Read with care
USDA's forecasts were issued at the time; Skyread's are backtest forecasts reconstructed in 2026 by a method chosen in 2026. The editions cover only three calendar years (2023, 2024 and 2025; none from October to December 2025, during the federal shutdown), and editions for the same year share one outcome, so this is evidence from three years, not from that many independent forecasts. ERS's historical forecast series is used: for 10 of the 112 category editions from January to August 2023 it differs from the numbers first published, by 0.1 point (0.3 for eggs in January 2023); ERS keeps no separate archive of later editions to check against.
| Category | Editions | Skyread error | USDA error | USDA range held |
|---|---|---|---|---|
| Grocery prices | 33 | 0.2 | 0.6 | 94% |
| Beef prices at the grocery store | 33 | 2.0 | 2.4 | 100% |
| Pork prices at the grocery store | 33 | 0.5 | 0.9 | 97% |
| Chicken and other poultry prices at the grocery store | 33 | 0.3 | 0.5 | 100% |
| Egg prices at the grocery store | 33 | 10.5 | 9.6 | 73% |
| Dairy prices at the grocery store | 33 | 0.5 | 0.7 | 100% |
| Fruit and vegetable prices at the grocery store | 33 | 0.3 | 0.3 | 100% |
| Bread, cereal and bakery prices at the grocery store | 33 | 0.3 | 0.8 | 91% |
| Prices of coffee, soft drinks and other nonalcoholic drinks at the grocery store | 33 | 0.4 | 1.0 | 94% |
| Sugar and sweets prices at the grocery store | 33 | 0.5 | 0.8 | 100% |
| Butter, oil and other fats prices at the grocery store | 33 | 0.6 | 1.4 | 91% |
| Fish and seafood prices at the grocery store | 33 | 0.7 | 0.7 | 100% |
| Prices of snacks, condiments, baby food and prepared foods at the grocery store | 33 | 0.2 | 0.6 | 88% |
| Restaurant menu prices | 33 | 0.2 | 0.3 | 94% |
Source: USDA ERS Food Price Outlook (agency_outlooks). USDA's numbers are quoted unchanged; Skyread is not affiliated with USDA.
EIA's Short-Term Energy Outlook (diesel, gasoline) is not compared yet: past editions are not stored.
Issued forecasts against what happened
No published outlook has reached its month yet. Results appear here as each forecast month is published, inside or outside its range.